ENHANCED LOCAL SECONDARY VOLTAGE CONTROL AND REACTIVE POWER BY WAY OFARTIFICIAL NEURON NETWORK

摘要

The improved local secondary voltage control by way of neuron networks is aimed at setting reference values of primary voltage controllers in power systems. It is based on autonomously operating local controllers located in all voltage controlled nodes. Typical of such a controller is that for its operation it requires only local voltage measurements of the controlled node and ranges of reactive power along linking lines to adjacent nodes. The controller is based on an artificial neuron network (UNO). By way of learning the information about the optimum operation of the power system (EES) and optimum voltage profile of the node is stored inside the UNO. The learning requires a learning group which includes values of reactive power throughput as well as actual and optimum voltage levels of the controlled node for a large number of operating conditions by the EES and which are calculated in advance with the help of an algorithm for the calculation of power throughput. Regarding the measured input data the secondary controller with the help of the UNO calculates the new value of reference voltage for the primary controller which adequately modifies the production of reactive power. The control process requires for its operation no tertiary control signal. The individual controllers operate completely autonomously therefore it is possible to apply them consecutively into the power system. Compared to the classical LSRN-UNO the key enhancement is determining the optimum UNO topology as well as several procedures for accelerating learning and increase of the convergence speed. All these procedures lead to higher effectiveness and quality o the controller operation with learnt UNO- s.